GigaScience Comparative transcriptomics of five high-altitude vertebrates and their low-altitude relatives --Manuscript Draft-Manuscript Number:
GIGA-D-17-00037R2
Full Title:
Comparative transcriptomics of five high-altitude vertebrates and their low-altitude relatives
Article Type:
Data Note
Funding Information:
National High Technology Research and Development Program of China (863 Program) (2013AA102502) National Natural Science Foundation of China (31402046) National Natural Science Foundation of China (31522055) National Natural Science Foundation of China (31601918) National Natural Science Foundation of China (31530073) National Natural Science Foundation of China (31472081) Science & Technology Support Program of Sichuan (2016NYZ0042) Youth Science Fund of Sichuan (2017JQ0011) China Postdoctoral Science Foundation (2015M572486) China Agriculture Research System (CARS-36) Program for Innovative Research Team of Sichuan Province (2015TD0012) Program for Pig Industry Technology System Innovation Team of Sichuan Province (SCCXTD-005) Project of Sichuan Education Department (15ZA0008) Project of Sichuan Education Department (15ZA0003) Project of Sichuan Education Department (16ZA0025) Project of Sichuan Education Department (16ZB0037) National Program for Support of Topnotch Young Professionals Young Scholars of the Yangtze River
Prof. Mingzhou Li
National Natural Science Foundation of China (31772576)
Prof. Mingzhou Li
Abstract:
Dr Qianzi Tang
Prof. Mingzhou Li
Dr Jideng Ma
Prof Xuewei Li
Prof. Mingzhou Li
Dr Yiren Gu
Dr Yiren Gu Dr Qianzi Tang Dr Yiren Gu Prof. Mingzhou Li
Dr Yiren Gu
Dr Xun Wang Dr Miaomiao Mai Dr Jideng Ma Dr An'an Jiang Prof. Mingzhou Li Prof. Mingzhou Li
Background: Species living at high altitude are subject to strong selective pressures due to inhospitable environments (e.g., hypoxia, low temperature, high solar radiation, and lack of biological production), making these species valuable models for comparative analyses of local adaptation. Studies that examined high-altitude Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
adaptation identified a vast array of rapidly evolving genes that characterize the dramatic phenotypic changes in high-altitude animals. However, how high-altitude environment shapes gene expression programs remains largely unknown. Findings: We generated a total of 910 Gb high-quality RNA-seq data for 180 samples derived from six tissues of five agriculturally important high-altitude vertebrates (Tibetan chicken, Tibetan pig, Tibetan sheep, Tibetan goat and yak), and their crossfertile relatives living in geographically neighboring low-altitude regions. Of these, ~75% reads could be aligned to their respective reference genomes, and on average ~70% of annotated protein coding genes in each organism showed FPKM expression values greater than 0.1. We observed a general concordance in topological relationships between the nucleotide alignments and gene expression-based trees. Tissue and species accounted for markedly more variance than altitude based on either the expression or the alternative splicing patterns. Cross-species clustering analyses showed a tissue-dominated pattern of gene expression, and a speciesdominated pattern for alternative splicing. We also identified numerous differentially expressed genes were potentially involved in phenotypic divergence shaped by highaltitude adaptation. Conclusions: This data serves as a valuable resource for examining the convergence and divergence of gene expression changes between species as they adapt or acclimatize to high-altitude environments. Keywords: high-altitude vertebrates, comparative transcriptomics, gene expression, alternative splicing Corresponding Author:
Mingzhou Li, Ph.D. Sichuan Agricultural University Chengdu, Sichuan CHINA
Corresponding Author Secondary Information: Corresponding Author's Institution:
Sichuan Agricultural University
Corresponding Author's Secondary Institution: First Author:
Qianzi Tang
First Author Secondary Information: Order of Authors:
Qianzi Tang Yiren Gu Xuming Zhou Long Jin Jiuqiang Guan Rui Liu Jing Li Keren Long Shilin Tian Tiandong Che Silu Hu Yan Liang Xuemei Yang Xuan Tao Zhijun Zhong Guosong Wang Xiaohui Chen
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Diyan Li Jideng Ma Xun Wang Miaomiao Mai An'an Jiang Xiaolin Luo Xuebin Lv Vadim N. Gladyshev Xuewei Li Mingzhou Li, Ph.D. Order of Authors Secondary Information: Response to Reviewers:
Reviewer 1: Comment 2-1 Your Series GSE93855 submission is labeled as unavailable until Jan 13 2020. Under the FAIR principles of publication in GigaScience, this must change if paper is accepted. Other datasets are also embargoed, far into 2018 or 2019. Response 2-1 Thank you for your reminder. We have released all of datasets related to our manuscript, inducing a total of 180 RNA-seq data deposited in the NCBI Gene Expression Omnibus (GEO) under accession numbers GSE93855, GSE77020 and GSE66242, as well as the 7 whole-genome sequencing data deposited in the NCBIsequence read archive (SRA) under accession number SRP096151. GSE93855: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE93855 GSE77020: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE77020 GSE66242: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE66242 SRP096151: https://trace.ncbi.nlm.nih.gov/Traces/sra/?study=SRP096151 Comment 2-2 The additional work performed to answer my reviewer comments is very good and shows interesting results. I don't have further major concerns, other than fixing the data availability issue above. Response 2-2 Thank you for your positive comments. Comment 2-3 The authors could consider revising Figure 4b and 4d, which are quite complex and difficult to see the relationships of all datasets to each other. The use of different color lines was not explained, and I tried to see why these were used, but was unable to see the logic. Response 2-3 Thank you for your valuable suggestion. In Fig. 4b and 4d, the vertical leading lines with different colors from the plotted points dropping to the xy-plane serve to improve the demonstration of group separation by tissue based on the first and second principal components. We added the explanation of the dropping vertical lines to the legends of Fig. 4b and 4d, “The vertical leading lines with different colors from the plotted points dropping to the xy-plane show the separation of points based on the first and second principal components”. To further promote the clarity of depiction for the principal component analysis (PCA) results, we added the two-dimensional PCA figures as the Supplementary Figs. S14-15 (accessible from Response_sup.pdf at: https://www.dropbox.com/s/sivc1djpvqzty2s/Response_sup.pdf?dl=0) Reviewer 2: Comment 3-1
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Thank you for addressing the previous points raised during the first reviewing stage. I believe that most points have been addressed in an appropriate way, and once again I thank you for your work and the data generated. Response 3-1 Thank you for your positive comments. Additional Information: Question
Response
Are you submitting this manuscript to a special series or article collection?
No
Experimental design and statistics
Yes
Full details of the experimental design and statistical methods used should be given in the Methods section, as detailed in our Minimum Standards Reporting Checklist. Information essential to interpreting the data presented should be made available in the figure legends.
Have you included all the information requested in your manuscript?
Resources
Yes
A description of all resources used, including antibodies, cell lines, animals and software tools, with enough information to allow them to be uniquely identified, should be included in the Methods section. Authors are strongly encouraged to cite Research Resource Identifiers (RRIDs) for antibodies, model organisms and tools, where possible.
Have you included the information requested as detailed in our Minimum Standards Reporting Checklist?
Availability of data and materials
Yes
All datasets and code on which the conclusions of the paper rely must be either included in your submission or deposited in publicly available repositories (where available and ethically appropriate), referencing such data using a unique identifier in the references and in the “Availability of Data and Materials” section of your manuscript.
Have you have met the above requirement as detailed in our Minimum Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Standards Reporting Checklist?
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Manuscript
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Click here to download Manuscript Manuscriptnew_HighLowAlt_ED.docx
1
Comparative transcriptomics of five high-altitude
2
vertebrates and their low-altitude relatives
3
Qianzi Tang1†, Yiren Gu2†*, Xuming Zhou3†, Long Jin1, Jiuqiang Guan4, Rui Liu1, Jing Li1,
4
Kereng Long1, Shilin Tian1, Tiandong Che1, Silu Hu1, Yan Liang2, Xuemei Yang2, Xuan
5
Tao2, Zhijun Zhong2, Guosong Wang1,5, Xiaohui Chen2, Diyan Li1, Jideng Ma1, Xun
6
Wang1, Miaomiao Mai1, An’an Jiang1, Xiaolin Luo4, Xuebin Lv2, Vadim N. Gladyshev3,
7
Xuewei Li1 and Mingzhou Li1*
8
1
9
Sichuan Agricultural University, Chengdu 611130, China;
Institute of Animal Genetics and Breeding, College of Animal Science and Technology,
10
2
11
Sichuan Animal Science Academy, Chengdu 610066, China
12
3
13
Medical School, Boston, Massachusetts, 02115 USA;
14
4
15
China;
16
5
17
USA.
18
† These
19
Corresponding authors. E-mail: Mingzhou Li:
[email protected], Yiren Gu:
20
[email protected].
Animal Breeding and Genetics Key Laboratory of Sichuan Province, Pig Science Institute,
Division of Genetics, Department of Medicine, Brigham and Women’s Hospital, Harvard
Yak Research Institute, Sichuan Academy of Grassland Science, Chengdu 610097,
Department of Animal Science, Texas A & M University, College Station, Texas, 77843
authors contributed equally to this work.
21 22 23 1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
24 25 26
Abstract
27
Background: Species living at high altitude are subject to strong selective
28
pressures due to inhospitable environments (e.g., hypoxia, low temperature,
29
high solar radiation, and lack of biological production), making these species
30
valuable models for comparative analyses of local adaptation. Studies that
31
examined high-altitude adaptation identified a vast array of rapidly evolving
32
genes that characterize the dramatic phenotypic changes in high-altitude
33
animals. However, how high-altitude environment shapes gene expression
34
programs remains largely unknown.
35
Findings: We generated a total of 910 Gb high-quality RNA-seq data for 180
36
samples derived from six tissues of five agriculturally important high-altitude
37
vertebrates (Tibetan chicken, Tibetan pig, Tibetan sheep, Tibetan goat and yak),
38
and their cross-fertile relatives living in geographically neighboring low-altitude
39
regions. Of these, ~75% reads could be aligned to their respective reference
40
genomes, and on average ~60% of annotated protein coding genes in each
41
organism showed FPKM expression values greater than 0.5. We observed a
42
general concordance in topological relationships between the nucleotide
43
alignments and gene expression-based trees. Tissue and species accounted
44
for markedly more variance than altitude based on either the expression or the
45
alternative splicing patterns. Cross-species clustering analyses showed a
46
tissue-dominated pattern of gene expression, and a species-dominated pattern
47
for alternative splicing. We also identified numerous differentially expressed
48
genes that could potentially be involved in phenotypic divergence shaped by 2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
49
high-altitude adaptation.
50
Conclusions: This data serves as a valuable resource for examining the
51
convergence and divergence of gene expression changes between species as
52
they adapt or acclimatize to high-altitude environments.
53
Keywords: high-altitude vertebrates, comparative transcriptomics, gene
54
expression, alternative splicing
55 56 57
Data description Transcriptome sequencing
58
Six tissues (heart, kidney, liver, lung, skeletal muscle and spleen) of three
59
unrelated adult females for each of five high-altitude vertebrates and their low-
60
altitude relatives were sampled (Fig. 1a and Supplementary Fig. S1). Animals
61
were sacrificed humanely to ameliorate suffering. All animals and samples used
62
in this study were collected according to the guidelines for the care and use of
63
experimental animals established by the Ministry of Agriculture of China. We
64
extracted total RNA, prepared libraries and sequenced the libraries on Illumina
65
HiSeq 2000 or 2500 platforms. We generated a total of ~909.6 Gb high-quality
66
RNA-seq data for 180 samples (~5.05 Gb per sample) of 30 individuals across
67
6 tissues (Supplementary Table S1).
68
Whole-genome re-sequencing
69
To compare the phylogeny derived from gene expression with the
70
phylogenetic relationships of the five high-altitude vertebrates and their low-
71
altitude relatives, we constructed the phylogenetic tree based on nucleotide
72
alignments. We extracted the unassembled reads from short-insert (500 bp)
73
libraries of a single yak [1] (NCBI-SRA: SRX103159 to SRX103161, and 3
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
74
SRX103175 and SRX103176), a Tibetan pig [2] (NCBI-SRA: SRX219342) and
75
a low-altitude Rongchang pig (NCBI-SRA: SRX1544519) [3] that were used for
76
de novo assemblies to roughly 10 × depth coverage. We also randomly
77
selected an individual of the cattle, low- and high-altitude chicken, goat and
78
sheep, and sequenced their whole genomes at ~10 × depth coverage (NCBI-
79
SRA: SRP096151). Genomic DNA was extracted from blood tissue of each
80
individual. Sequencing was performed on the Illumina X Ten platform, and a
81
total
82
(Supplementary Table S2).
of
198.64
Gb
of
paired-end
DNA sequence
was
generated
83 84 85
Data analysis Data filtering
86
To avoid reads with artificial bias, we removed the following type of reads: (a)
87
Reads with ≥ 10% unidentified nucleotides (N);
(b) Reads with > 10 nt aligned
88
to the adapter, allowing ≤ 10% mismatches; (c) Reads with > 50% bases having
89
phred quality < 5.
90
Identification of single-copy orthologous genes
91
Single-copy orthologous genes across five reference genomes, i.e. chicken
92
(Galgal4) [4], pig (Suscrofa 10.2) [5], cattle (UMD3.1) [6], goat (CHIR_1.0) [7]
93
and sheep (Oar_v3.1) [8] were determined using a EnsemblCompara
94
GeneTrees method [9] (Supplementary Fig. S2, Supplementary Methods)
95
[9].
96
Construction of phylogenetic tree based on nucleotide alignments
97
High-quality re-sequencing data were mapped to their respective reference
4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
98
genomes using BWA software, version 0.7.7 (BWA, RRID:SCR_010910) [10],
99
reads with mapping quality > 0 were retained and potential PCR duplication
100
cases were removed. For each individual, ~97.01% of reads were mapped to
101
~97.40% (at least 1 × depth coverage) or ~91.86% (at least 4 × depth coverage)
102
of
103
Single nucleotide variations (SNVs) and insertion and deletions (InDels) were
104
further detected by following GATK’s best practice, version 3.3-0 (GATK,
105
RRID:SCR_001876) [11]. We substituted SNVs and InDels identified in our
106
study in the coding DNA sequences (CDS) of the respective reference
107
genomes. Single copy orthologues with substituted CDS of the five vertebrates
108
were applied to Treebest [12] and generating the neighbor-joining tree (Fig. 1b).
109
Analyses of gene expression
the
reference
genome
assemblies
(Supplementary
Table
S2).
110
High-quality RNA-seq reads were mapped to their respective reference
111
genomes using Tophat version 2.0.11 (TopHat, RRID:SCR_013035) [13].
112
Cufflinks version 2.2.1 (Cufflinks, RRID:SCR_014597) [14] was applied to
113
quantify gene expression and obtain FPKM expression values. We generated
114
abundance files by applying Cuffquant (part of Cufflinks) to read mapping
115
results. Log2-transformed values of (FPKM + 1) for genes with >0.5 FPKM in
116
over 80% of the samples were used for subsequent analyses.
117
Pearson’s correlations were calculated across six samples from low- and
118
high-altitudes populations within each group of specific tissue and animals;
119
among pairwise comparisons of five animals within each of the six tissues; and
120
among pairwise comparisons of six tissues within each of the five animals.
121
Principal Variance Component Analysis (PVCA) was carried out using R
122
package pvca [15]. Neighbor-joining expression-based trees were generated 5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
123
according to distance matrices composed of pairwise (1-Spearman’s
124
correlations) implemented in the R package ape [16]. Reproducibility of
125
branching patterns was estimated by bootstrapping genes, that is, the single
126
copy orthologues were randomly sampled with replacement 100 times. The
127
fractions of replicate trees that share the branching patterns of the original tree
128
constructed were marked by distinct node colors in the figure.
129
We generated abundance files by applying Cuffquant (part of Cufflinks) to
130
read mapping results, and further applied abundance files to Cuffdiff (part of
131
Cufflinks) to detect DEGs between population pairs from distinct altitudes
132
within each group of specific tissue and species. Genes with FDR-adjusted p-
133
values ≤ 0.05 were detected as DEGs.
134
Genes were converted to human orthologs, and assessed by DAVID
135
(DAVID, RRID:SCR_001881) [17] webserver for functional enrichment in GO
136
(Gene Ontology) terms consisting of molecular function (MF) and biological
137
process (BP) as well as the KEGG (KEGG, RRID:SCR_012773) pathways
138
and InterPro (InterPro, RRID:SCR_006695) databases (Benjamini adjusted p-
139
values ≤ 0.05).
140
Analyses of alternative splicing
141
Single-copy orthologous exons were identified by finding annotated exons that
142
overlapped with the query exonic region in a multiple alignment of 99 vertebrate
143
genomes including human genome (hg38) from the UCSC genome browser
144
[18]. Exon groups with multiple overlapping exons in any species were
145
excluded. Each internal exon in every annotated transcript was taken as an 6
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
146
“cassette” exon. Each “cassett” alternative splicing (AS) is composed of three
147
exons: C1, A and C2, where A is alternative exon, C1 the 5’ alternative exon,
148
C2 the 3’ alternative exon. For each species and read length k, we generated
149
all non-redundant constitutive and alternative junction sequences for the
150
following RNA-seq alignments. The junction sequences were constructed by
151
retrieving k-8 bp from each of the two exons making up the junction, and when
152
the exon length is smaller than k-8, the whole sequence of the exon is retrieved.
153
This ensures that there is at least 8 bp overlap between the mapped reads and
154
each of the two junction exons.
155
We then estimated the effective number of uniquely mappable positions of
156
the junctions. We extracted L-k+1 (L being the junction length) k-mers from
157
each junction and mapped such k-mers back to the reference genome allowing
158
up to two mismatches. Those k-mers that failed to align were further mapped
159
to the non-redundant junctions. The number of k-mers that could uniquely align
160
to a junction was counted and deemed as the effective number of uniquely
161
mappable positions for the junction.
162
For each sample, RNA-seq reads were first aligned to the reference genome
163
allowing up to two mismatches, and the unaligned reads were further mapped
164
to the non-redundant junctions. Uniquely mapped reads for each junction were
165
counted, and multiplied by the ratio between the maximum number of mappable
166
positions (i.e. k-15) to the effective number of uniquely mappable positions for
167
the junction.
168
The “percent-spliced in” (PSI) values for each internal exon was defined as
169
PSI = 100 × average (#C1A, #AC2) / (#C1C2 + average(#C1A, #AC2)), here
170
#C1A, #AC2 and #C1C2 are the normalized read counts for the associated 7
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
171
junctions. Exons were taken as alternative in a sample if 5≤PSI≤95. We also
172
defined “high-confidence” PSI levels as those that meet the following criteria:
173
*max(min(#C1A, #AC2), #C1C2) ≥ 5 AND min(#C1A, #AC2) + #C1C2 ≥ 10
174
*|log2(#C1A / #AC2)| ≤ 1 OR max(#C1A, #AC2) < #C1C2
175
For cross-species analyses, we included exons with single-copy orthologues
176
in all species, PSI values in all samples, and confidently alternative spliced in
177
at least one of the samples.
178 179 180 181 182
Findings Data summary
183
We generated a total of ~909.6 Gb high-quality RNA-seq data, of which ~676.6
184
Gb (~74.6%) reads could reliably aligned to their respective reference genomes
185
(Supplementary Fig. S3 and Table S1). We found that on average 61.2%
186
annotated protein coding genes in each genome had FPKM expression values
187
greater than 0.5 (Supplementary Fig. S4 and Table S3).
188
Concordance in the tree topology based on nucleotide sequence
189
alignments and gene expression data
190
Nucleotide alignments-based phylogenetic relationships of these high-altitude
191
vertebrates and their low-altitude relatives matched the
192
morphological species groupings and the known history of population formation
193
(Fig. 1b). The gene expression-based tree based 4,746 transcribed single-copy
194
orthologous genes (66.61% of 7125) for each tissue showed a highly consistent
8
established
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
195
topology to the nucleotide sequence alignment-based phylogeny (Fig. 2,
196
Supplementary Methods) [9]: mammals were mainly divided into omnivore
197
(pig) and ruminant (goat, sheep and yak/cattle); within the ruminant cluster, the
198
two caprinae (goat and sheep) were closer to each other than the bovinae
199
(yak/cattle). This observation lends supports the idea that gene expression
200
changes evolve together with genetic variation over evolutionary time, resulting
201
in lower expression divergence between more closely species [19].
202
Distinctly transcriptomic characteristics between gene expression and
203
alternative splicing
204
Through comparison of expression levels of 4,746 transcribed single-copy
205
orthologous genes (Supplementary Fig. S2) and alternative splicing patterns
206
(reflected by PSI values) of 2,783 orthologous exons shared by the five
207
vertebrates genomes, we observed a tissue-dominated clustering pattern of
208
gene expression, but a species-dominated clustering pattern of alternative
209
splicing [20, 21].
210
For gene expression, there were critical biological differences among tissues
211
(Pearson’s r = 0.67 and weighted average proportion variance = 0.36), followed
212
by species (Pearson’s r = 0.75, weighted average proportion variance = 0.22)
213
and local adaptation (Pearson’s r = 0.95 and weighted average proportion
214
variance = 0.019) (Fig. 3a and Supplementary Fig. S5). By contrast, for
215
alternative splicing, the differences among species (Pearson’s r = 0.64 and
216
weighted average proportion variance = 0.30) were higher than among tissues
217
(Pearson’s r = 0.78 and weighted average proportion variance = 0.075),
218
followed by between high- and low-altitude animals (Pearson’s r = 0.84 and
219
weighted average proportion variance = 0.021) (Fig. 3b and Supplementary 9
220 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
Figure S6).
221
Unsupervised clustering (Figs. 4a and 4c) and principal components
222
analysis (PCA) (Figs. 4b and 4d and Supplementary Figs. S7 and S8) both
223
recapitulated the distinctly transcriptomic characteristics between gene
224
expression and alternative splicing. Tissue-dominated clustering of gene
225
expression indicated that in general tissues possess conserved gene
226
expression signatures and suggested that conserved gene expression
227
differences underlie tissue identity in mammals. On the other hand, greater
228
prominence of species-dominated clustering of alternative splicing suggested
229
that exon splicing is more often affected by species-specific changes in cis-
230
regulatory elements and/or trans-acting factors than gene expression [20, 21].
231
Notably, tissue-dominated clustering patterns of gene expression further
232
revealed that the cluster of striated muscle (heart and skeletal muscle) and the
233
cluster of vessel-rich tissues (lung and spleen) were closer to each other than
234
the cluster of metabolic tissues (kidney and liver), followed by the distinct
235
clusters of bird (chicken) and mammals according to the evolutionary distance
236
(Figs. 4a and 4b). Notably, tissues of birds (chickens) formed a distinct cluster,
237
rather than with their mammalian counterparts, which indicates that divergence
238
in gene expression among these species started to surpass that between
239
different tissues around when birds diverged from mammals (approximately
240
300 million years ago) (Figs. 4a and 4b).
241
Gene expression plasticity to a high-altitude environment
242
To exclude the impact of prominence of tissues-dominated clustering of gene
243
expression, so as to comprehensively present transcriptomic differences
10
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
244
involved in high-altitude response based on whole annotated genes of their
245
respective genome assembly instead of the single-copy orthologous, we
246
measured the pairwise difference of gene expression between the high-altitude
247
populations and their low-altitude relatives within each tissue for each
248
vertebrate.
249
We identified ~1,423 DEGs between 30 low- versus high-altitude pairs (177
250
DEGs in muscle of chickens to 3,853 DEGs in kidney of sheep) (Table 1).
251
Notably, among five pairs of vertebrate, the highly-diverged yak and cattle [1]
252
exhibited the highest number of DEG (~2,005) across six tissues. Among six
253
tissues, the highly aerobic kidney [22] exhibited the highest number of DEGs
254
(~2,097) across five pairs of vertebrates.
255
Expectedly, the more closely related vertebrates (Fig. 1) shared more DE
256
genes (Supplementary Figs. S9–10 and Additional File 3). Compared with
257
shared DE genes among mammals, especially between the two closely related
258
members of Caprinae (goat and sheep), the birds (chickens) exhibited
259
significantly fewer shared DE genes with mammals (Wilcoxon rank sum test,
260
P
